1. Field of the Invention
The present invention relates to compositions and methods for the selection of nucleic acids and polypeptides.
2. Description of the Related Art
Ligand-receptor interactions are of interest for many reasons, from elucidating basic biological site recognition mechanisms to drug screening and rational drug design. It has been possible for many years to drive in vitro evolution of nucleic acids by selecting molecules out of large populations that preferentially bind to a selected target, then amplifying and mutating them for subsequent re-selection (Tuerk and Gold, Science 249:505 (1990), herein incorporated by reference).
The ability to perform the selection process with proteins would be extremely useful. This would permit in vitro design and production of proteins that bind specifically to chosen ligands. The use of proteins, as compared to nucleic acids, is particularly advantageous because the twenty diverse amino acid side chains in proteins have far more binding possibilities than the four similar chains in nucleic acid side. Further, many biologically and medically relevant ligands bind proteins.
Both nucleic acid and protein evolution methods require access to a large and highly varied population of test molecules, a way to select members of the population that exhibit the desired properties, and the ability to reproduce the selected molecules with mutated variations to obtain another large population for subsequent selection.
Prior attempts to develop a protein evolution method were primarily limited by the inability of the proteins to reproduce themselves and the inability to link a polypeptide's encoding mRNA with the translated product. Additionally, the generation of large peptide libraries and screening methods have, until recently, required that the process have an in vivo expression step. Examples include yeast two- or three-hybrid, yeast display and phage display methods (Fields and Song, Nature 340:245 (1989); Licitra and Liu, PNAS 93:12817 (1996); Boder and Wittrup, Nat Biotechnol 15:553 (1997); and Scott and Smith, Science 249:386 (1990)). In vivo methods suffer from various disadvantages, including a limited library size and cumbersome screening steps. Additionally, undesired selective pressures can be placed on the generation of variants by cellular constraints of the host.
In vitro methods have been developed more recently, using prokaryotic and eukaryotic in vitro translation systems, such as ribosome display. (Mattheakis et al., PNAS 91:9022 (1994); Hanes and Plückthun, PNAS 94:4937 (1997); Jermutus et al., Current Opinion in Biotechnology 9:534 (1998), all herein incorporated by reference). These methods link the protein and its encoding mRNA with the ribosome and the entire complex is screened against a ligand of choice. Potential disadvantages of this method include the large size of the ribosome, which could interfere with the screening of the attached, and relatively tiny, protein.
In 1997, two groups of workers developed an in vitro method of attaching a protein to its coding sequence during translation by using the ribosomal peptidyl transferase with puromycin attached to a linker DNA (Szostak et al., International Patent Publication WO 98/31700; Roberts and Szostak PNAS 94:12297 (1997); Nemoto et al., FEBS Letters 414:405 (1997), all herein incorporated by reference). Once the coding sequence and peptides are linked, the peptides are exposed to a selected ligand. Selection or binding of the peptide by the ligand also selects the attached coding sequence, which can then be reproduced by standard means. Both Roberts and Szostak and Nemoto et al. used the technique of attaching a puromycin molecule to the 3′ end of a coding sequence by a DNA linker or other non-translatable chain. Puromycin is a tRNA acceptor stem analog which accepts the nascent peptide chain under the action of the ribosomal peptidyl transferase and binds it stably and irreversibly, thereby halting translation. These methods suffer from certain disadvantages. For example, the coding sequence encoding each peptide must be known and be modified both initially and between each selection. Thus, the methods of Roberts and Nemoto cannot be used to select native unknown mRNAs. Further, the modification of the coding sequence adds several steps to the process. Finally, the attached puromycin on the linker molecules may compete in the translation reaction with the native tRNAs for the A site on the ribosome reading its coding sequence or a nearby ribosome, and could thus “poison” the translation process, just as would unattached puromycin in the translation reaction solution. Inadvertent interactions between puromycin and ribosomes could result in two kinds of reaction non-specificity: prematurely shortened proteins and proteins attached to the wrong message. There are reports in the prior art that indicate that the avidity of the A site and the peptidyl transferase for the puromycin may be modulated by Mg++ concentration (Roberts, Curr. Opin. Chem. Biol. 3:268 (1999), herein incorporated by reference). Although Mg++ concentration may be titrated to control for the first kind of non-specificity, i.e. premature termination of translation, it will not affect the second type, i.e. inaccurate mRNA-protein linkage.
Thus, a need exists for an in vitro nucleic acid-based protein evolution system which does not necessarily require initial knowledge of the nucleic acid's sequence or repeated chemical modification of the nucleic acids, and which can accurately link a mRNA to its protein.